BaMBNet:A Blur-Aware Multi-Branch Network for Dual-Pixel Defocus Deblurring

作     者：Pengwei Liang Junjun Jiang Xianming Liu Jiayi Ma Pengwei Liang;Junjun Jiang;Xianming Liu;Jiayi Ma

作者机构：School of Computer Science and TechnologyHarbin Institute of TechnologyHarbin 150001China Electronic Information SchoolWuhan UniversityWuhan 430072China IEEE 

出 版 物：《IEEE/CAA Journal of Automatica Sinica》 (自动化学报（英文版）)

年 卷 期：2022年第9卷第5期

页      面：878-892页

核心收录：

学科分类：08[工学] 080203[工学-机械设计及理论] 0802[工学-机械工程] 0811[工学-控制科学与工程] 

基　　金：supported by the National Natural Science Foundation of China (61971165, 61922027, 61773295) in part by the Fundamental Research Funds for the Central Universities (FRFCU5710050119) the Natural Science Foundation of Heilongjiang Province(YQ2020F004) the Chinese Association for Artificial Intelligence(CAAI)-Huawei Mind Spore Open Fund 

主　　题：Blur kernel convolutional neural networks(CNNs) defocus deblurring dual-pixel(DP)data meta-learning 

摘      要：Reducing the defocus blur that arises from the finite aperture size and short exposure time is an essential problem in computational *** is very challenging because the blur kernel is spatially varying and difficult to estimate by traditional *** to its great breakthrough in low-level tasks,convolutional neural networks(CNNs)have been introdu-ced to the defocus deblurring problem and achieved significant ***,previous methods apply the same learned kernel for different regions of the defocus blurred images,thus it is difficult to handle nonuniform blurred *** this end,this study designs a novel blur-aware multi-branch network(Ba-MBNet),in which different regions are treated *** particular,we estimate the blur amounts of different regions by the internal geometric constraint of the dual-pixel(DP)data,which measures the defocus disparity between the left and right *** on the assumption that different image regions with different blur amounts have different deblurring difficulties,we leverage different networks with different capacities to treat different image ***,we introduce a meta-learning defocus mask generation algorithm to assign each pixel to a proper *** this way,we can expect to maintain the information of the clear regions well while recovering the missing details of the blurred *** quantitative and qualitative experiments demonstrate that our BaMBNet outperforms the state-of-the-art(SOTA)*** the dual-pixel defocus deblurring(DPD)-blur dataset,the proposed BaMBNet achieves 1.20 dB gain over the previous SOTA method in term of peak signal-to-noise ratio(PSNR)and reduces learnable parameters by 85%.The details of the code and dataset are available at https://***/junjun-jiang/BaMBNet.